See https://github.com/era7bio/salmon16spb/pull/9#issuecomment-346541632 for the context. BLAST fails badly on empty reads. So we shouldn't pass them to BLAST and record their IDs in a separate output. Together with #130 we will have 3 types of discarded outputs from BLAST:
empty reads (IDs)
reads with no hits (FASTA/Q)
hits that don't pass the after-BLAST filter (?)
This particular filtering could be also done on split step. I'll think about it.
See https://github.com/era7bio/salmon16spb/pull/9#issuecomment-346541632 for the context. BLAST fails badly on empty reads. So we shouldn't pass them to BLAST and record their IDs in a separate output. Together with #130 we will have 3 types of discarded outputs from BLAST:
This particular filtering could be also done on split step. I'll think about it.